20200730 developer call notes

[hammer]

Pre-notes

Process and tools

• API discussions on the main repo? sgkit-requirements#7
• Move forum to pystatgen.org? topic
• Grabbing a semaphore before doing work: assign an issue to yourself?
• All non-trivial PRs should have issues
• Note out to PySnpTools team (Carl and Danilo) to try to join forces

Open PRs

• read_vcfzarr #40: blocked on Alistair’s feedback
• Sphinx #51: ready to go?
• CLI Plugins #57: click plugins useful? Do we need this now?
• Display genotypes #58: Tom needs to handle big data
• REGENIE #66 #69
• Docs describing Genotype Call XArray #78: do we want notebooks?

Open Issues

• How can we avoid fixed-length string truncation by dask/zarr? #71
• How should we test jit-compiled functions? #77
• What disadvantages exist for splitting multi-allelic sites and doing a bi-allelic analysis? #21
• We need to make sure Xarray variable name parts (between slashes) do not match dimension names #62
• Small documentation conventions:
  • Type annotations in docstrings #72
  • Docstring linter #70
  • Delimited test names (“__”) #75
  • Is there anything we don’t put a period at the end of, even if it’s not a sentence?
• Enforce style across pystatgen repos #65
  • How to “have a single repo with proper separation of modules to prevent dependency hell”?

Notes

Attendees

• Jeff Hammerbacher
• Alistair Miles
• Eric Czech
• Rafal Wojdyla

Discussion

• 10 am ET good or should we move it later?
• How to best get Quansight integrated with the development process?
  • AM thinks there’s more work to be done in understanding requirements
  • Let’s discuss where our workloads intersect?
  • Popgen stuff is pretty separate
  • JH: TW will work w/ Quansight on core operation
• Opinions on pystatgen vs. pydata-statgen vs. sgkit
  • No one has strong opinions
• Grab a semaphore before working on something
  • Let’s try to make an issue before there’s a PR
  • EC: non-trivial, but unanticipated
  • AM: time to capture and communicate our roadmap?
• read_vcfzarr #40
  • AM: want to be able to read in Ag1000G
• CLI Plugins #57
  • EC: working w/ UK Biobank so many corner cases to make a CLI useful
• REGENIE #66 #69
  • EC formulated some conventions that could use review
  • RW notes branches need to be in the main repo to merge PRs into PRs
  • RW notes it’s good to have a convention for naming branches
  • AM: best way into new functionality is a worked example
• Docs describing Genotype Call XArray #78
  • AM: what is the right way of documenting examples? Cf. Sphinx Gallery.
  • AM: MyBinder/Google Colab live links are really nice.
  • JH: we’re early so let’s keep out of docs
  • JH: we have precedent with validation/ of having in the repo
  • AM: would like a structured tutorial, cf. Xarray Tutorial
• How can we avoid fixed-length string truncation by dask/zarr? #71
• How should we test jit-compiled functions? #77
  • RW: pytest markers
  • AM: tricky when working with code coverage tools
• Enforce style across pystatgen repos #65
  • RW: use extras_require?